A Service Oriented Approach for Distributed Data Mediation on the Grid

Seamless integrated access to data stored in globally distributed databases has become a major challenge in many scientific disciplines. In this paper, we describe a Grid-based service-oriented infrastructure that tackles this challenge through the provisioning of virtual data sources realized by means of distributed data mediation services. Virtual data sources offer to the user a single integrated view of distributed heterogeneous databases and information sources while hiding the details of data location, data formats, and access mechanisms of the underlying physical data sources. Personalized views of data sources, tailor-made for specific usage scenarios, may be offered by different virtual data sources. Virtual data sources rely on flexible mediation techniques and utilize distributed query processing to optimize complex data integration scenarios. Distributed data mediation services have been realized on top of standard Grid and Web Services technologies including OGSA-DAI and OGSA-DQP. The generic data service infrastructure described in this paper is being utilized in the context of the European @neurIST project, which develops an advanced service-oriented Grid infrastructure for the management and treatment of multi-factorial diseases. Details of distributed data mediation and query processing are presented in the context of an experimental scenario integrating clinical data bases.